As readers here know, the Columbus Dispatch poll of political contests in Ohio, which according to its authors had a stellar record in calling elections up until 2004, had a disastrous 2005 general election (with issues and no candidates), a very weak 2006 primary, and a similarly weak 2006 general election.
The paper’s poll did very well in the 2008 presidential election, predicting that Barack Obama would defeat John McCain by 6 points; Obama won by 4.6%.
This year, the Dispatch passed on the primary.
Final Poll: 49-47-4
Election Results: 49.3-46.7-4.0
Difference: Basically none
Portman-Fisher Other —
Final Poll: 56-40-4
Election Results: 57.2-39.0-3.8
Difference: Underestimated GOP margin by 2 points (within the poll’s 2.3 point margin of error)
Final Poll: 44-50-6
Election Results: 47.8-46.0-6.2
Difference: Underestimated GOP margin by 8 points
Final Poll: 47-46-6
Election Results: 50.5-44.6-4.9
Difference: Underestimated GOP margin by 5 points
Final Poll: 52-42-6
Election Results: 54.0-41.1-4.9
Difference: Underestimated GOP margin by 3 points
Final Poll: 51-42-7
Election Results: 54.9-40.2-4.9
Difference: Underestimated GOP margin by 6 points
Overall: In six races, the poll underestimated the Republican candidates’ margins by an average of 4 points, compared to over 12 points in 2006.
Assessment: Clearly very much improved. The Dispatch has every right to crow about having nailed the governor’s race and staying within the margin of error in the U.S. Senate contest.
Down-ticket, the the results were not quite as impressive. The paper missed on its prediction that Rich Cordray would defeat Mike DeWine. It underestimated the GOP margins in the four races involved by an average margin of 5.5 points. Still, that’s less than half of its underestimate in 2006.
That’s much better. But, until shown otherwise, future down-ticket statewide GOP candidates can take some comfort in the Dispatch’s pre-election poll results if they are slightly trailing their Democratic opponents, and can feel pretty confident of their chances if the paper is showing that they are ahead.